The Eh vs. Geu Problem

by Strang Burton and Martina Wiltschko


In Canadian English, you can use the tag eh to request confirmation for a proposition that you have some (but not necessarily complete) reason to believe, as in You got your hair cut, eh? Other English dialects, and many other languages, have related ‘confirmational’ words—but often with subtly different systematic relations to context.

This squib explores the difference between Canadian eh and a closely equivalent tag in Upper Austrian German (hence UAG): geu (pronounced /gɔɪ/). We argue that the tags are differentiated by a ‘previous reason to believe’ confirmational parameter, active in UAG but not in English

The Problem: eh vs. geu

In many contexts both eh and geu are well-formed, with the same force (i.e. requesting confirmation for a belief). For example:

Context A: eh and geu both good

cartoon talking about Alan, Mark mentions he goes to the gym

Cartoon showing Mary running into Alan on an empty street
Mary says You go to the gym', then can add eh or geu in German equivalent

But in other contexts where eh is good, geu is clearly ungrammatical (though an alternative, introduced later in this squib, is allowed). For example:

Context B: eh good, but geu bad

Mary is thinking of Alan, wondering how he spends his time

Mary sees Alan outside the gym
Mary says 'So you go to the gym now', she can add 'eh' but not in German 'geu'

So what’s going on? How do speakers use eh and geu, such that they occur as equivalents in some contexts, but not in others?

The Proposal: the R2B-PR2B Parameter

We propose the following solution to the eh vs. geu problem:

  • Eh and geu are the same, in the following respect: both tags require you to have some reason to believe (hence R2B), before you request confirmation.

  • Eh and geu differ, in the following respect: with geu (but not eh) your R2B must be established prior to the conversation. That is, for geu you need not only an R2B, but a previous R2B (PR2B)

More generally, our claim is that the confirmational systems in Canadian English and UAG differ: UAG confirmationals (but not their English equivalents) are sensitive to when you obtained the R2B.


Our account predicts the distribution for Canadian eh and UAG geu shown below.

chart summarzing predictions.  Ey is good with any RTB, geu is only good if the RTB is revious

These predictions account for the Eh-Geu problem as follows:

  • In Context A (where Mary heard previously about Alan going to the gym) Mary’s R2B is previous to the conversation. Geu is good, because the R2B is previous, and eh is also good just because there is some R2B.

  • In Context B (where Mary only grasps that Alan goes to the gym as the conversation starts) Mary’s R2B is NOT previous to the conversation. Eh is good, because there is still some R2B,but geu is bad, because this R2B is not previous.

Because our account proposes a general +/- PRTB distinction within UAG, we also predict the possibility of a confirmational which is the converse of geu: that is, a confirmational which is bad when the R2B is previous, and good when it is not. Such a word does exist in UAG: the adverb leicht.

Chart now adds 'leicht', which is only good with an RTB that is -PREV'

As our account predicts, leicht (as in Du tranierst leicht im Fitness Studio?) is bad in Context A above (where the R2B is +PREV), and good in Context B (where the R2B is -PREV).

We will not illustrate the predictions for the other contexts, but quick further investigation would reveal the remaining prediction: that all three confirmationals are bad in contexts with no RTB at all.


With respect to the distinctions discussed above (RTB an PRTB), American huh behaves the same as Canadaian eh, so that US huh is also good in both Context A and Context B. However, Canadian eh and US huh also differ in other ways, as discussed here.


Martina Wiltschko, Linguistics Dept. University of British Columbia

Strang Burton, Stolo Shxweli Halq’emeylem Language Program, Stolo Nation, and Linguistics Dept. University of British Columbia

Suggested citation:
Burton, Strang, and Martina Wiltschko. 2015. The Eh vs. Geu Problem. In Short ’schrift for Alan Prince, compiled by Eric Baković.

 back to Squibs